Food inspection apparatus, method of inspecting food, and method of learning in identification unit of food inspection apparatus

Abstract

A food inspection apparatus includes: a conveyance unit; a light irradiation unit; an unit capturing an image of an inspection object A; a wavelength emphasis emphasizing a foreign substance specific wavelength characteristic of a foreign substance F from light having a wavelength of 300 nm to 1100 nm by using a first and/or second optical filter or a wavelength-specific light source; and an identification processing device identifying the foreign substance and including: a lightening unit normalizing the captured image with 256 or less of gradations into lightened data; and an identification unit having been provided with deep-learning of an identification processing of the foreign substance specific wavelength from the lightened data and identifying the foreign substance or a good item S in line from the lightened data obtained through capturing the image during conveyance of the inspection object.

The invention claimed is: 
     
       1. A food inspection apparatus comprising:
 a conveyor configured to convey a food as an inspection object to be confirmed that no foreign substance is mixed; 
 a light source configured to irradiate the inspection object with light; 
 an imager configured to capture an image of the inspection object; 
 a wavelength emphasizer including at least one of a first optical filter attached to the light source or the imager, a second optical filter attached to an auxiliary light source, and a wavelength-specific light source, and configured to emphasize a characteristic wavelength reflected by at least one of the food and the foreign substance; and 
 a processor configured to identify at least one of the food and the foreign substance, 
 wherein the wavelength emphasizer emphasizes at least one of a foreign substance specific wavelength characteristic of the foreign substance and a good item specific wavelength characteristic of a good item of the food from light having a wavelength in a range of 300 nm to 1100 nm, 
 wherein the processor is further configured to: 
 normalize the image captured by the imager with 256 or less of gradations of light intensity to obtain lightened data; and 
 be provided in advance with deep-learning to identify the foreign substance specific wavelength from the lightened data through a convolutional neural network for color reconstruction based on luminance by using the lightened data of plural image sheets, and 
 wherein the processor identifies the foreign substance or the good item in line from the lightened data obtained through capturing the image during conveyance of the food by the conveyor. 
 
     
     
       2. The food inspection apparatus according to  claim 1 ,
 wherein the wavelength emphasizer emphasizes at least one foreign substance specific wavelength and at least one good item specific wavelength, and 
 wherein the processor identifies the good item in line from the lightened data obtained through capturing the image during conveyance of the food by the conveyor. 
 
     
     
       3. The food inspection apparatus according to  claim 2 ,
 wherein the wavelength emphasizer emphasizes two wavelengths of a first foreign substance specific wavelength and a second foreign substance specific wavelength as the foreign substance specific wavelength, and 
 wherein the processor processes the image captured by the imager as a pseudo RGB image. 
 
     
     
       4. A method of learning for a processor of a food inspection apparatus, comprising:
 extracting, from a training image obtained by capturing an image of a food to be confirmed that no foreign substance is mixed, a characteristic wavelength reflected by the foreign substance as a foreign substance specific wavelength and a characteristic wavelength reflected by a good item of the food as a good item specific wavelength; 
 preparing a plurality of pseudo RGB training images at least including the foreign substance specific wavelength and the good item specific wavelength; 
 normalizing the pseudo RGB training images with 256 or less of gradations of light intensity and lightening the images to obtain lightened learning data; and 
 providing the processor with deep-learning to identify the foreign substance specific wavelength of the good item specific wavelength from the lightened learning data through a convolutional neural network for color reconstruction based on luminance. 
 
     
     
       5. A food inspection apparatus comprising a processor provided with learning by the method of learning according to  claim 4 . 
     
     
       6. A method of inspecting a food, confirming that no foreign substance is mixed in the food as an inspection object by using a food inspection apparatus according to  claim 1 . 
     
     
       7. A method of inspecting a food, confirming that no foreign substance is mixed in the food as an inspection object by using a food inspection apparatus according to  claim 5 .